CCE encoding questions (regarding video modes)

[nautilus7]

I use cce sp2 to encode my dvds. I prefer 1-pass vbr (q factor) to create a .vaf file and multipass vbr (2 passes) after. Is this correct?

Wouldn't be the same if i used multipass vbr (3 passes, creating the .vaf file in the 1st pass)?

What exactly is te Q factor and how do i set it correctly?

[Guest]

Yes, you've done something wrong. You've failed to follow our forum rules, specifically, rule 9: use a proper thread title. Please fix it. Hit Edit and then Go Advanced on your first post. Thank you.

http://forum.doom9.org/forum-rules.htm

[PhillipWyllie]

No this isn't correct. Q factor is only available for 1 pass VBR( don't confuse this with 1st pass VBR). Q factor( I think) is equivalent to constant quality(where the quantisation scales are varied). If you want to do VBR, in my opinion, select 1st pass VBR, then set your target file size/ average bitrate. After the generation of the vaf file select mulitipass VBR, set how many passes,make sure "create new" for the video information file isn't checked. You can play around with vaf file via VBR setting to set exactly how the bitrate is distributed. If 1st pass VBR isn't selected first when multipass VBR is selected, then CCE will initially do an extra pass to generate the vaf file, then carry on with the other passes.

[Boulder]

If you intend to do a multipass VBR encode, there's really no need to do an OPV encode. In fact, it can make the next pass perform worse if the Q value you set in OPV isn't correct.

[PhillipWyllie]

You can't do a multipass q-factor based encode as you can't set the q value when you select multipass. Also the vaf file doesn't take into account the q-factor(this just determines the quantisation scales).

[Boulder]

If you use a Q value that would result in an average bitrate that is very much different from the target, the next pass (i.e. first VBR pass) would not be as good as it can be. When you first predict the Q value so that it'll hit close to the target, the results should be good. I've never tested if it's any better than a regular multipass so I can't say for sure.

I'm quite sure that there is a reason why CCE usually targets the vaf file creation pass so that the average bitrate is very close to the actual target (even though it doesn't create any m2v file). I've seen it undersize only when you need to enable adaptive quant matrices and have never seen it really oversize.

[nautilus7]

Let's see if i got it right...

1st pass vbr (vaf) + N passes multi pass vbr = N+1 passes in multi vbr when i set create new vaf file

Not use one pass vbr (Q) + multi vbr together.

[45tripp]

yes, that's right
it'll create a vaf file if there isn't one, whether you tick create new or not.
1st pass of vbr option outputs both vaf and mpv file in a single pass.

[nautilus7]

Do you prefer multipass vbr, or Q related passes?

What speed do you get in cce? My brand new core 2 6600 gets ~5,5, but only reaches 70% cpu usage. Why? I use a very simple avs script to frameserve and i set dgdecode to use skall's idct.

[45tripp]

2 pass vbr always for me. (2+1vaf)

Speed is good. It's the fastest.
Depends on source and filtering.
On my P4 dog, say 35 fps (per pass that is)

[vwpassion]

@nautilus: I can't seem to go above 70% cpu usage either with my E6400 Core2Duo. Not sure where the problem is. Frameserving a simple .avs script as well, using dgdecode - even tried using different idct's. But still only 70% cpu usage @ 5,5 - 6,5 speed.
(Sorry for offtopic)

[Boulder]

More CPU usage doesn't necessarily mean better performance. In fact, it can be quite the opposite.

[manono]

If you guys are worried about not using your fancy CPUs to their fullest, then you can multi-thread AviSynth, which is pretty much guaranteed to get the useage above 90%:

http://forum.doom9.org/showthread.php?t=94996
http://avisynth.org/mediawiki/MT_support_page

But, as Boulder says, you may not always get better performance. Depending on what I'm doing, I get anywhere from a 50% boost in encoding speed, to a small decrease in encoding speed.